Essays on Foreign Aid and Macro-Economic Performance of Sub-Saharan African Countries
by
Omar Saleh
M.A., Simon Fraser University M.B.A., University of Toledo
Post Baccalaureate Diploma in Economics, Simon Fraser University B.Sc., Lebanese American University
A dissertation submitted to the Faculty of Graduate Studies in partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY
in the Department of Economics
© Omar Saleh, 2019 University of Victoria
All rights reserved. This dissertation may not be reproduced in whole or in part, by photocopy or other means, without the permission of the author.
ii Supervisory Committee
Essays on Foreign Aid and Macro-Economic Performance of Sub-Saharan African Countries
by
Omar Saleh
M.A., Simon Fraser University M.B.A., University of Toledo
Post Baccalaureate Diploma in Economics, Simon Fraser University B.Sc., Lebanese American University
Supervisory Committee
Dr. Alok Kumar, Department of Economics Supervisor
Dr. Merwan Engineer, Department of Economics Departmental Member
Dr. Sorin Rizeanu, Gustavson School of Business Outside Member
iii Abstract
Foreign aid is a major flow of income into sub-Saharan African (SSA) countries, averaging roughly 12% of GDP over the last four decades. Yet, SSA countries are characterized by very low per capita output, low human capital attainment, and widespread poverty. This dissertation investigates the macroeconomic and welfare effects of foreign aid to SSA countries. The empirical part of the dissertation studies 22 SSA countries, and uses a cointegrated vector autoregressive analysis (CVAR). This methodology identifies long-run effects without imposing strong statistical priors. I introduce tradable and non-tradable sectors into the analysis to determine if the so-called “Dutch Disease” is the reason for the plight of SSA countries. “Dutch Disease” occurs when a positive shock to foreign aid perversely reduces GDP, by decreasing the relative price of tradable to non-tradable goods, thus reducing the size of the non-tradable sector. While I find that aid reduces GDP in eight countries, this result is inconsistent with the “Dutch Disease” as it is not accompanied by large relative price changes. The analysis controls for a number of country-specific characteristics including extraordinary events. Overall, I find non-positive impacts of foreign aid on GDP and the tradable sector, with a few exceptions. I also consider the reverse causal channel and test whether country-specific macroeconomic variables drive foreign aid flows. I find that GDP, tradable output, and tradable and non-tradable goods prices do affect the amount of aid a country receives in 15 countries. These variables have no impact on foreign aid (aid is considered as weakly exogenous) in six countries.
The theoretical part of the dissertation develops two dynamic stochastic general equilibrium — real business cycle — (DSGE-RBC) models to analyze the effects of foreign aid on human capital investment and the business cycle. The distinguishing feature of the models is to embed a human capital investment in a small open economy model of Mendoza (1991). The first model considers one-sector DSGE model, which is followed by two-sector (tradable and non-tradable) DSGE model. Both models distinguish between physical and human capital investment and allow for labor-leisure choice. In the analysis, labor supply and time spent studying or acquiring skills are optimally chosen. The models are calibrated to match the key features of the Kenyan economy. In both models, a positive aid shock initially has a negative impact on labor supply and output. However, the shock subsequently has a positive effect on physical and human capital investment, and time spent studying. This is due to a positive income effect from the shock. A rise in foreign aid increases consumption; consumption smoothing across periods raises physical and human capital investment, labor productivity, and output. I also find that reducing the volatility of aid has
iv a significant positive effect on human capital investment and welfare. Policymakers should focus on reducing the volatility of foreign aid and not solely concentrate on the average level of aid. The analysis of the two-sector DSGE-RBC model incorporates the role for the “Dutch Disease” mechanism. Consistent with the “Dutch Disease”, I find that a shock to foreign aid appreciates the relative price of non-tradable goods that causes the factors of production to reallocate from the tradable sector to the non-tradable sector, leading to a decline in GDP and the tradable output. Finding the “Dutch Disease” result here is not necessarily at odds with the CVAR estimation results as the DSGE-RBC simulation is a short-run analysis and the CVAR estimation is a long-run analysis.
v Table of Contents Supervisory Committee……….ii Abstract………..iii Table of Contents…...v List of Tables...…...x List of Figures………...…...xiii Acknowledgements………..……….xv Dedications………...…xvi Chapter 1 Introduction………..…….1 Chapter 2 Kenya……….7 2.1 Introduction………..………..7 2.2 Agricultural Sector……….7
2.3 Manufacturing, Mining and Quarrying Sectors ………...………….8
2.4 Other Industries………..9
2.5 Service Sector.………….………..9
2.6 GDP’s Composition and the Relative Price of Non-tradable Goods………...9
2.7 Schooling System in Kenya……….10
2.8 Foreign Aid………..10
Chapter 3 Have Sub-Saharan African Countries Caught the Dutch Disease from Foreign Aid? A Cointegrated VAR Analysis………...12
3.1 Introduction………..12
3.2 Literature Review……….15
3.2.1 Foreign Aid and the GDP’s Composition………..………16
3.2.2 Foreign Aid and the “Dutch Disease”………...……….17
3.3 Data Sources and Stationarity Tests...……….19
3.3.1 Data Sources………..19
3.3.1.1 Human Capital Data………20
3.3.2 Data Transformation………..20
3.3.3 Stationarity tests……...………...………...21
vi
3.3.3.2 Unit Root Tests with Structural Breaks…...………22
3.4 Econometric Approach: The Cointegrated VAR (CVAR) Framework………...23
3.4.1 Methodological Motivation……….………..23
3.4.2 The Cointegrated VAR Model…………..……….24
3.4.3 The Common Trends Representation………….………...………28
3.4.4 Model Specifications……….………...……….31
3.4.4.1 Deterministic Components………..……….……….…..31
3.4.4.2 Dummy Variables………...….………32
3.4.5 Misspecification Tests………...34
3.4.6 Rank Determination……….…..34
3.4.7 The CVAR Stability Parameter Tests………37
3.5 Causal Links between Foreign Aid and the Variables in the SSA Countries………..38
3.6 The Long-Run Impact of Aid on the Variables……….…..39
3.7 Robustness Checking………...42
3.7.1 Aid Effectiveness/Ineffectiveness………..…43
3.7.2 The Impact of Aid on the GDP’s Structure………..………….44
3.8 The Impact of GDP and the Tradable Output on Foreign Aid Flows…………...…45
3.9 Summary and Conclusion………47
Chapter 4 Foreign Aid, Human Capital Investment, and Business Cycle………..……60
4.1 Introduction………..60
4.2 Literature Review……….62
4.2.1 Schooling and Business Cycle………...62
4.2.2 Cyclicality of Foreign Aid……….63
4.2.3 Effects of Foreign Aid………...64
4.2.4 Effects of Remittances………...…65
4.2.5 Foreign Aid and Welfare………...65
4.2.6- Similarities and Differences between the Present Model and Previous Models………..65
4.3- Data Sources………..……….66
vii
4.4.1 Competitive Equilibrium………...…………71
4.4.2 Equilibrium Dynamics………..……….71
4.5 DSGE-RBC Calibration and Results……….………..73
4.5.1 Parameters……….….73
4.5.1 Calibration Results……….………74
4.6 Simulation Results………..……….76
4.6.1 Short Term Impact from Temporary Changes in Shocks…………...…………..76
4.6.1.1 Foreign Aid Shock………..76
4.6.1.2 Public Human Capital Shock……….……….……77
4.7 Sensitivity Analyses………...…………..78
4.7.1 Permanent Changes in Foreign Aid Levels………..78
4.7.1.1 Foreign Aid Shock at Different Aid Levels………...79
4.7.2 Foreign Aid Shock at Different Persistent Levels………80
4.7.3 Effects of Changing the Volatility of Aid: Mean-Preserving Spread.…………..82
4.7.4 Forecast Error Variance Decomposition………...83
4.7.5 Correlated Shocks versus Uncorrelated Shocks…...……….85
4.8 Conclusion………...85
Chapter 5 Human Capital Investment, Foreign Aid, and the “Dutch Disease”………....….99
5.1 Introduction……….….99
5.2 Literature Review………...101
5.3- Data Sources……….102
5.4 The Theoretical DSGE-RBC Model……….……….103
5.4.1 Competitive Equilibrium……….…………110
5.4.2 Equilibrium Dynamics……….110
5.5 DSGE-RBC: Calibration and Results………..…………..113
5.5.1 Parameters………...……….113
5.5.1 Calibration Results………...…………114
5.6 Simulation Results……….115
5.6.1- Short Term Impact from Temporary Changes in Shocks…………...……..….116
viii
5.6.1.2 Public Human Capital Shock……….………...………117
5.7 Sensitivity Analyses……….…..118
5.7.1 Permanent Changes in Foreign Aid Levels………118
5.7.2 Foreign Aid Shock at Different Persistent Levels………..…………120
5.7.3 Effects of Changing the Volatility of Aid: Mean-Preserving Spread.…….…..122
5.7.4 Forecast Error Variance Decomposition………..…………..123
5.7.5 Correlated Shocks versus Uncorrelated Shocks…...………..124
5.8- The Vector Error Correction Model (VECM)……….………...………..125
5.8.1- Long-run Result………..126
5.8.1.1- Long-run Exclusion Test ………..………...126
5.8.2- The VECM Results ……….………127
5.8.2.1- The VECM Impulse Response Functions………128
5.9- Conclusion………...………...……..129
Chapter 6 Conclusion………...…..142
References…………...……….145
Appendix A: The CVAR Model……….………158
A.1 SSA Countries not Included in the Study and Dummy Variables………158
A.2 Aid, “Dutch Disease” Mechanism, and Exports Competitiveness………...163
A.3 Data Definitions and Measurements……….167
A.4 The CVAR Model...171
A.4.1 The Common Trends Representation……….…..171
A.4.2 The Unrestricted MA Representation………...172
A.5 Misspecification Tests...172
A.5.1 Determination of the Lag Length………..172
A.5.2 Residual Cross-Correlations Tests………..…..173
A.5.3 Residual Autocorrelations Tests………...174
A.5.4 Residual Heteroscedasticity Tests……….175
A.5.5 Normality Tests……….176
ix
A.6 Tradable Sector: Change in Composition...180
Appendix B: One-sector DSGE Model...183
B.1 The Representative Agent’s Optimization Problem...183
B.2 Firm’s Optimization Problem...185
B.3 The Equivalent Social Planner’s Problem.……….……...185
B.4 Steady State...188
B.5 Balanced Growth Path...189
B.6 Estimates of the Share of Households Human Capital Investment...191
B.7 World Bank Database Definition...192
B.8 One-sided HP Filter versus Two-sided HP Filter………..194
B.9 Technology and Government Expenditure Shocks………...196
B.9.1 Technology Shock (TFP Shock)………..………...196
B.9.2 Government Expenditure Shock……….197
Appendix C: Two-sector DSGE Model...200
C.1 The Representative Agent’s Optimization Problem...200
C.2 Firms’ Optimization Problems...204
C.3 The Equivalent Social Planner’s Problem...206
C.4 Steady State...210
C.5 Balanced Growth Path...211
C.6 Technology Parameters’ Estimates...214
C.7 Technology and Government Expenditure Shocks………...217
C.7.1 Technology Shocks (TFP Shocks)………..217
C.7.2 Government Expenditure Shock……….218
C.8 Stationarity and rank Tests ……….………..221
x Lists of Tables
Table 3.1A: Countries, sample period, aid, tradable output, exports and imports as percentage of
GDP and components of the tradable sector (Source: WDI, 2017)………...49
Table 3.1B: Composition of tradable Sector (Source: WDI, 2017)………50
Table 3.2A: Perron test for exogenous structural break………..51
Table 3.2B: Zivot-Andrews test for endogenous structural break………..52
Table 3.2C: Unit-root test results with two structural breaks……….…52
Table 3.3: Ranking test………...53
Table 3.4: Testable hypotheses of causal links between aid and the variables……….……….….54
Table 3.5: Long-run impact of aid (rank based on trace test)……….…………55
Table 3.6: The number of Case I-IV SSA countries according to sign and statistical significance of the effect of foreign aid on the variables………...56
Table 3.7: Long-run impact of aid (rank based on trace test)……….…57
Table 3.8: A sensitivity analysis of the long-run impact of foreign aid on GDP, the tradable and non-tradable outputs under two economic hypothesis……….………..58
Table 3.9: The long-run impact of aid on GDP structure……….………..58
Table 3.10: Long-run impact of the variables (rank based on trace test)………..………….…….59
Table 3.11: Comparison between the impacts of the variables’ shocks………...………..59
Table 4.1: Calibration for the benchmark model……….………...87
Table 4.2A: Data and the references of the benchmark model (Means)……….…..…..88
Table 4.2B: Data and the references of the benchmark model (Standard deviations)………..…..88
Table 4.2C: Data and the references of the benchmark model (Correlations with GDP)…...……88
Table 4.2D: Data and the references of benchmark model (Correlation with aid)……….…89
Table 4.2E: Data and the references of the benchmark model (Correlation with public human capital)……….……89
Table 4.2F: Data and the references of the benchmark model (Correlation with government Expenditure)………...89
Table 4.3A: Sensitivity analysis of the benchmark model (Means)……….….….90
Table 4.3B: Sensitivity analysis of the benchmark model (Standard deviations)…………...…...90
Table 4.4: Impacts of changing aid’s volatility………..91
xi
Table 4.6: The benchmark model and the shocks………..….92
Table 4.7: Volatilities and correlations of GDP and foreign aid………93
Table 5.1: Calibration for the benchmark model………..131
Table 5.2A: Data and the references of the benchmark model (Means)………..132
Table 5.2B: Data and the references of the benchmark model (Standard deviations)………….132
Table 5.2C: Data and the references of the benchmark model (Correlations with GDP)………133
Table 5.2D: Data and the references of the benchmark model (Correlations with aid)………...133
Table 5.2E: Data and the references of the benchmark model (Correlations with public human capital)………...133
Table 5.2F: Data and the references of the benchmark model (Correlations with government expenditure)………..133
Table 5.3: Foreign aid sensitivity simulations………..134
Table 5.4: Impacts of changing aid’s volatility………....135
Table 5.5: Variance decomposition………..135
Table 5.6: The benchmark model and the shocks……….136
Table 5.7: VECM Results…………...…….……….137
Table A1: SSA countries not included in the study……….158
Table A2: Ease of doing business rank……….159
Table A3: Dummy variables and reasons for inclusion………..……..160
Table A4: Residual cross-correlation with aid………..174
Table A5: Autocorrelation tests (LM tests)…....……….….175
Table A6: ARCH tests………..176
Table A7.1: Normality tests (P-Value)……….……178
Table A7.2: Skewness and Kurtosis Values……….179
Table A8: Long-run impact of aid (Tradable = Agriculture + Industry)………..180
Table A9: Comparison of long-run impact of aid when the composition of tradable sector changes (GDP and tradable output)……….181
Table A10: Comparison of long-run impact of aid when the composition of tradable sector changes (tradable and non-tradable outputs)……….…182
xii
Table B2: Distribution of households by type of migrant………...…….192
Table B3.1: Kenyan data moments (Standard deviations)………..……..195
Table B3.2: Kenyan data moments (Correlation with GDP)………..……..195
Table B3.3: Kenyan data moments (Correlation with aid)………...195
Table B3.4: Kenyan data moments (Correlation with public human capital)………..195
Table B3.5: Kenyan data moments (Correlation with government expenditure)……….195
Table C1: Estimates of labor and capital share in Kenyan agricultural output……….215
Table C2: Estimates of labor and capital share in Kenyan industrial and services sectors……..216
Table C3.1: Augmented Dickey-Fuller unit root test………..….221
Table C3.2: Phillips-Perron unit root test……….221
Table C3.3: KPSS Unit Root Test Results………221
Table C3.4: Perron test for exogenous structural break………..……..222
Table C3.5: Zivot-Andrews test for endogenous Structural Break……….………..222
xiii Lists of Figures
Figure 1.1: Aid to developing countries (Source: OECD, 2017)………3 Figure 1.2: Net per capita official developments assistance to SSA countries
(Source: WDI, 2017)………..4 Figure 2.1: Non-tradable-to-tradable output ratio and the relative price of non-tradable goods (Source: Author’s calculation based on WDI, 2017)……….10 Figure 2.2: Kenyan aid-to-GDP ratio 1970-2014 (Source: WDI, 2017)………..11 Figure 2.3: Aid to education share by educational levels (Averages 2002–2014;
Sources: OECD, 2017)………..11 Figure 4.1: Impulse responses to a positive one-standard foreign aid shock……….….94 Figure 4.2: Impulse responses to a positive one-standard public human capital shock…………...95 Figure 4.3: Sensitivity analysis for aid shock impact on selected variables at different aid-to-GDP levels……….96 Figure 4.4: Sensitivity analysis for aid impact on selected variables at different aid shock
persistence levels………..97 Figure 4.5: Impulse responses to a one-standard deviation of different shocks……….…….98 Figure 5.1: Impulse responses to a positive one-standard-deviation foreign aid shock…………..138 Figure 5.2: Impulse responses to a positive one-standard-deviation public human capital shock..139 Figure 5.3: Sensitivity analysis for aid impact on selected variables at different aid shock
persistence levels………140 Figure 5.4: The VECM impulse response functions………….………...………….141 Figure B1: Impulse responses to a positive one-standard TFP shock………..…….198 Figure B2: Impulse responses to a positive one-standard government expenditure shock…….…199
xiv Figure C1: Impulse responses to a positive one-standard-deviation TFP shock………..….219 Figure C2: Impulse responses to a positive one-standard-deviation government expenditure shock………...220
xv Acknowledgements
I would like to express my sincere gratitude to my supervisor Dr. Alok Kumar for his enormous support, abundant supervision, and encouragement over the course of my Ph.D. research. In addition, I am very thankful to my advisor Dr. Merwan Engineer who has been an outstanding mentor. Dr. Kumar’s and Dr. Engineer’s flexibility, understanding, and their valuable suggestions were of immense value without which I could not have finished my Ph.D. dissertation. I am grateful to both of them for their trust in me. I’d also like to thank Dr. Sorin Rizeanu for serving on my Doctoral committee and for his support.
I would like to express my gratitude to the faculty and staff of the Department of Economics. In particular, I’d like to thank Dr. Graham Voss, Dr. Fatemeh Mokhtarzaddeh, Dr. Felix Pretis, Dr. Ke Xu, Dr. Kenneth Stewart, Dr. Marco Cozzi, Dr. Chris Auld, Dr. Paul Schure, and Dr. Maggie Jones for their valuable comments and suggestions.
I am deeply and eternally grateful to my dear wife, Nahla. Throughout my studies, she has been a pillar of support. Without her loving and unconditional support, completing my Ph.D. dissertation would have been impossible. My love goes to my little daughters, Ikram and Syrine, who give me the determination to succeed.
Finally, I would like to express my deep appreciation and thankfulness to my mother, Ikram, Uncle Akram, and to my sisters, Zeinab and Lara, who with their love, support, and encouragement enabled me to persevere through the challenges of my Ph.D. studies.
xvi Dedications
Chapter 1
Introduction
The people of sub-Saharan Africa (SSA) are the poorest in the world: about 42% of the estimated 910 million people in SSA, in 2013, lived below the World Bank’s extreme poverty line of US $1.90 a day1. This extreme poverty persists despite SSA countries receiving more than a trillion
dollars2 in foreign aid since 1960. The ratio of foreign aid (official development assistance) to GDP
in SSA countries has averaged 12% from 1970 to 2014, averaging around 8% in 2014. It is an open question of whether foreign aid to SSA countries has been effective in promoting economic development and growth.
Education is the cornerstone of economic development. It improves the efficiency of physical capital utilization and accelerates an economy’s technological progression. It stimulates development and improves people’s lives by increasing efficiency and fostering democracy (Barro, 1997; 2001), thus creating better conditions for quality governance, improving the health care system, and reducing inequality (Aghion et al., 1999). In particular, Barro (1991), Lucas (1988), Mankiw et al. (1992), and Sen (1999) emphasize the role of education and human capital accumulation, in the growth process from a macroeconomic perspective.
At the beginning of the 21st century, the United Nations (UN) had included universal access to
education as the second of its Millennium Development Goals (MDGs). Yet, many parts of sub-Saharan Africa are still struggling to achieve primary school completion rates above 68%. The average number of schooling years completed in SSA was only five years in 2010. In 2016, UNESCO3 stated that SSA countries had the highest rates of children and youth out of school: 20%
of children aged six to 11 years, 33% of youth aged 12 to 14 years, and nearly 60% of the youth
1 Calculations based on the database from World Development Indicators (2018). “Poverty headcount ratio
at $1.90 a day is the percentage of the population living on less than $1.90 a day at 2011 international prices” (WDI, 2018).
2 In 2010 constant US dollars, OECD database.
2 aged 15 to 17 were not attending school. Guarcello et al. (2015) examine 25 developing countries, including eight SSA countries, and find that the majority of out-of-school children are usually engaged in child labor4. Education is one of the key components that can help in reducing child
labor.
The UN’s Sustainable Development Goals (SDGs) for the period 2015–2030 emphasize universal secondary education as well as access to job-skills training at higher education levels (Sustainable Development Solutions Networks Thematic Group on Early Childhood Development, Education and Transition to Work, 2014). By 2030, more than 600 million children in the world need to be enrolled in school to achieve the SDG goal of education for all. According to the UNESCO Global Education Monitoring (GEM) Report (2015), the world needs $39 billion annually to provide 12 years of quality education for all. According to the UNICEF Report (2015), 46 low-income countries, mostly located in sub-Saharan Africa, would need $26 billion yearly to achieve this goal. Poor educational attainment in SSA is a result of numerous factors including insufficient educational infrastructure which can be traced to insufficient resources available in these countries (Birdsall et al., 2001). Foreign aid can be aimed to bridge the resource gap between the required investment and the available resources, and can play a crucial role in meeting the SDGs.
Foreign aid is a major source of income for SSA countries. If well targeted, it can help recipient countries progress along their developmental paths (Sachs, 2005). From 1971 to 2016, total aid to education increased in real terms5 from $10.3 billion to $12.2 billion - a 20% increase. However,
over the same period, total aid disbursed to developing countries has increased by 400% and reached $176 billion in 2016 (See Figure 1.1a). During this period, aid to education as a share of total aid fell from around 23% to 7%. Figure 1.1b shows that aid to education as a share of total aid is decreasing, which means that achieving the SDG goal of education for all is likely to become more difficult. Figure 1.1c shows the distribution of aid to education from 2002 to 2016, whereas Figure 1.1d illustrates the averages for the period. On average, aid to tertiary education constituted 37% of total aid to education followed by aid to primary education which constituted around 30%, while aid to secondary education only constituted 10% of total aid to education.
4 32% of children aged seven to 11 years and 23% of children aged 12 to 14 years are working. 5 Constant 2016 US$.
3 Figure 1.1: Aid to developing countries (Source: OECD 2017)
Figure 1.2 illustrates the behavior of the net official development assistance (ODA) per capita to SSA countries, for the last four decades. During the 1970–1994 period, the net ODA per capita was increasing; however, between 1994 and 2000, aid per capita declined due to structural adjustment programs adopted by the World Bank and the International Monetary Fund (IMF). Since 2001, the net ODA per capita began to increase, but by 2006 it started to decline. In 2005, the G86 held their
meeting in Gleneagles, Scotland, and agreed to double the net ODA per capita to Africa by 2010, and keep it at that level thereafter. Unfortunately, the net ODA per capita started to fall shortly after the Gleneagles commitment. The decline in the net ODA per capita raises an important policy question: How can the SSA countries achieve the SDG goal of universal education?
4 Figure 1.2: Net per capita official developments assistance to SSA countries (Source: WDI, 2017)
20 30 40 50 60 70 80 1970 1975 1980 1985 1990 1995 2000 2005 2010 AID_PER_CAPITA While foreign aid may increase human capital investment by increasing resources; it may also have adverse effects. Foreign aid may appreciate the relative price of non-tradable goods (i.e. real exchange rate), leading to the reallocation of resources away from the tradable sector. The appreciation of the relative price of non-tradable goods reduces the competitiveness of exports, especially the manufactured goods (Corden, 1984), and has an adverse effect on the productive efficiency (Doucouliagos and Paldam, 2009). This so-called “Dutch Disease” has received considerable attention in the literature. Foreign aid is argued to have a negative impact on income and growth. Aid may also have adverse effects on economic sectors that are most in need of development such as the human-capital intensive tradable sector; this may reduce the return to education and decrease human capital investment. The manufacturing and services sectors are skill-intensive; skills must be continuously improved to maintain high profits and returns to skills which can be achieved by investing in education (Gylfason, 2001).
Another important question is whether the variability and uncertainty of foreign aid have detrimental welfare effects that undo the possible welfare benefits of foreign aid. The volatility in the flow of foreign aid arises largely from irregular aid disbursement patterns and donors conditionality (Pallage and Robe, 2001; Bulir and Hamann, 2003, 2008; Bulir and Lane, 2004). Unexpected aid shortfalls can lead recipient governments to reduce investment in human and physical capital, whereas aid windfalls increase government consumption (Celasun and Walliser, 2008). Fluctuations in the amount of foreign aid to the low development countries may have significant implications on the volatility of total output, consumption, and investment in those countries. The volatility of aid may negatively affect investment and welfare.
5 This dissertation provides a comprehensive analysis of the long-run impact of foreign aid to SSA countries. In particular, I investigate the impact of foreign aid on GDP as well as its impact on the tradable and non-tradable outputs and prices. First, I start with an empirical investigation of 22 SSA countries, using a cointegrated vector autoregressive (CVAR) model. Then, I provide a theoretical analysis, by using dynamic stochastic general equilibrium — real business cycle — (DSGE-RBC) models. A one-sector DSGE-RBC model is used to investigate the short-run impact of foreign aid on output, private human capital investment, and time devoted to acquire new skills. A two-sector DSGE-RBC model is developed to incorporate the role of the “Dutch Disease” mechanism and its effect on the GDP’s structure, in addition to total output and human capital accumulation.
Both DSGE-RBC models are calibrated to match the salient features of the Kenyan economy for the 1970–2014 period. Because I study Kenya in greater details, Chapter 2 provides a description of the key features of the Kenyan economy.
Chapter 3 develops the empirical CVAR analysis of the long-run impact of aid flows on GDP and its composition for 22 sub-Saharan African countries. I find that aid has a persistent negative long-run impact on GDP in eight countries, a positive impact in three countries, and no effect on the rest of the countries. Foreign aid also appears to have diverse impacts on the GDP’s composition. I find that aid has a positive impact on the tradable output in four countries and on the non-tradable output in six countries. On the other hand, aid flows have persistent negative impacts on the tradable and non-tradable sectors in 16 and five countries, respectively. I do not find that the negative impacts on the tradable sector are caused by the mechanism of a decline in the relative price of tradable to non-tradable goods. Thus, the analysis does not support that foreign aid to SSA countries causes the “Dutch Disease”. For Kenya, a foreign aid shock has a persistent negative impact on GDP and the tradable output. However, foreign aid has only a transitory impact on tradable and non-tradable goods prices.
In Chapter 4, I develop a one-sector DSGE-RBC model of the Kenyan economy. I find that a positive shock to aid has business cycle implications. Initially, it lowers labor supply and output. However, subsequently, it has a positive effect on output which is due to the positive income effect. A rise in foreign aid increases consumption; consumption smoothing across periods raises physical and human capital investment, time spent studying, and labor productivity. The one-sector DSGE-RBC model finds that reducing the volatility of foreign aid has a significant positive effect on human capital investment and welfare. Policymakers should focus on reducing the volatility of foreign aid and not only on the average level of aid as foreign aid is highly volatile in SSA countries.
6 Chapter 5 extends the one-sector DSGE-RBC model in Chapter 4 into a two-sector DSGE-RBC model, by splitting the total output into tradable and non-tradable sectors. This extension incorporates the “Dutch Disease” mechanism into the model. Here, I find results similar to those in Chapter 4 concerning output, private human capital investment, study time, labor supply, physical capital investment, and consumption. In addition, the two-sector DSGE-RBC model suggests that a positive foreign aid shock leads to an appreciation of the relative price of non-tradable goods. The appreciation of the relative price of non-tradable goods leads to a reallocation of resources from the tradable to non-tradable sector, causing the tradable sector to shrink and the total output to drop. In addition, results show that decreasing the volatility of aid increases private human capital investment and study hours. Finally, a vector error correction model (VECM) analysis provides results that are consistent with the dynamics of the two-sector DSGE-RBC model.
The rest of this dissertation is organized as follows: Chapter 2 introduces Kenya, the country I choose to calibrate my DSGE-RBC models. The empirical part of the dissertation is presented in Chapter 3. Chapter 4 develops the one-sector DSGE-RBC model. In Chapter 5, I develop the two-sector DSGE-RBC model. The conclusion is discussed in Chapter 6.
7
Chapter 2
Kenya
2.1- Introduction
Kenya is located in East Africa and covers an area of 581,309 𝑘𝑘𝑘𝑘2and has an estimated population of 50 million people in 2017; per capita income in Kenya was estimated at $3000 (PPP). Kenya is considered a typical aid recipient country where the aid-to-GDP ratio averaged around 6% for the period 1970–2014 period. I have chosen Kenya for the following reasons. First, it is an aid-dependent country. Second, Kenya’s aid receipts as a share of GDP have been about half those of other sub-Saharan countries (See Table 3.1A), so doubling or tripling foreign aid is more applicable than elsewhere. Third, Kenya has produced higher-quality macroeconomic data than many other SSA countries.
Kenya gained its independence from the United Kingdom on December 12th, 1962. Since
independence, Kenya has been a relatively stable country in sub-Sahara Africa with only one military coup attempt on August 1st, 1982. However, a political and humanitarian crisis occurred
between December 2007 and February 2008 that led to chaos where 1,500 people were killed and about 600,000 people were displaced. Below, I discuss the structure of the Kenyan economy. 2.2- Agricultural Sector
Agriculture in Kenya is an important sector. It employed around 61.1% of the total employed labor force in 2005 (KILM, 2005). In 2014, the agricultural sector including forestry and fishing contributed 28% of the Kenyan value-added GDP, a decline from around 31% in 1970 (WDI, 2017). For the period 1970–2014, the agricultural sector grew at an average growth rate of 3.2% which is the same rate of population growth in Kenya. The agricultural sector is characterized by a poor performance which has negative implications on poverty levels as well as on the standard of living of the Kenyan people.
8 Around 48% of Kenyan land (274500 𝑘𝑘𝑘𝑘2) is classified as agricultural land; however, 16% of Kenyan land (91500 𝑘𝑘𝑘𝑘2) is considered as high to medium agricultural potential, while the rest is arid and semi-arid land (ASAL) with low agricultural potential. About 11% of the Kenyan population lives in the ASAL region and the rest lives in high to medium agricultural potential land regions. Most of the agricultural products are produced by small farms. There are about three million small-scale farms7, where 81% of these farms are less than 20,000 𝑘𝑘2 each8. The
small-scale farms contribute between 70-75% of the country’s total value of the agricultural output and about 85% of the total employment in the agricultural sector; the rest is produced by the large-scale farms9 (Odhiambo et al., 2004). The main agricultural products are maize, rice, sugar cane, coffee,
tea, raw milk and horticultural.
The usage of agriculture machinery (tractors) is not intensive in Kenya. In 2002, tractors per 100 𝑘𝑘𝑘𝑘2 of arable land were 25, up from 20 in 1970. This number is 10% of that of the USA (WDI,
2017). The production in the large-scale farms relies heavily on techniques that are capital intensive (i.e. mechanised harvesting, tractors, etc…) and are more productive than the small-scale farms which are characterized by high labor intensity and the usage of traditional technologies (i.e. ox-drawn carts for plowing) (Odihiambo et al., 2004).
2.3- Manufacturing, Mining and Quarrying Sectors
The manufacturing sector in Kenya contributed 10% of the value added GDP in 2014. For the period 1970–2014, the manufacturing sector averaged about 10.5% of the total output, peaking to 12.8% in 2007 (WDI, 2017). On average the manufacturing sector grew by 5% for the same period. In 2005, only 4.2% of the total employed labor force was employed in the manufacturing sector (KILM, 2005).
There are many obstacles facing the manufacturing sector in Kenya. For instance, private and public investments in the manufacturing sector are low and are not enough to enable the sector to take off. Public infrastructure investments are essential to reduce transaction costs, while private investments in technology and research and development (R&D) are essential to increase efficiency and productivity. Both types of investment are required for the manufacturing companies in Kenya to compete internationally (Bigsten et al., 2010). However, R&D expenditure as a percentage of GDP in Kenya is very low compared to developed countries. In 2010, Kenya spent 1% of its GDP
7 Areas of the farms range from 5,000 to 100,000 𝑘𝑘2.
8 Kenya's Agricultural Produce. Retrieved from:
http://www.kenyabrussels.com/index.php?menu=6&leftmenu=88&page=89
9 on R&D (WDI, 2017). In reality, Kenyan manufacturing firms rely heavily on foreign technology and foreign R&D. The location of Kenya at the equator makes it far from the countries that are technology leaders. This remoteness is a disadvantage relative to countries that are closer to the leaders (Keller, 2002). The main manufacturing products in Kenya are the following: food manufacturing, textiles, footwear, leather, rubber and plastic, petroleum, industrial chemicals, paints, furniture and fixtures, soft drinks, cement, and metal products.
The mining and quarrying sector contributed less than 0.8% of the Kenyan GDP in 2014 (KNBS, 2015). The main minerals produced are: soda ash, limestone, gold, salt, and fossil fuel.
2.4- Other Industries
This sector includes telecommunication, construction, electricity, gas, and water. This sector is heavily capital intensive and contributed 7.25% of the total value added GDP in 2014. This sector grew on average by 5% for the 1970–2014 period (WDI, 2017). In 2005, 2.6% of the total employed labor force was working in other industries (KILM, 2005).
2.5- Service Sector
The service sector in Kenya is the largest sector. Service sector’s contribution represented 48% of the Kenyan economy in 2014 (WDI, 2017). In 2005, 32.2% of the employed labor force was employed in the service sector (KILM, 2005). The service sector as a share of GDP has increased from 42% in 1970 to 48% in 2014 and grew on average by 5%. The service sector includes the following services: wholesale and retail trade, hotels and restaurants, storage, transport and communications, insurance, financial intermediaries, real estate, business services, government, education and healthcare, community social and other personal services.
2.6- GDP’s Composition and the Relative Price of Non-tradable Goods
The tradable output is composed of following sectors: agricultural, forestry, hunting, fishing and manufacturing sectors. The rest of the GDP is assumed to be a non-tradable output. The tradable output-to-GDP ratio has declined from 41% in 1970 to 37% in 2014. Figure 2.1 below shows the non-tradable-to-tradable output ratio and the relative price of non-tradable goods. Consistent with the “Dutch Disease” phenomenon, the movements in the non-tradable-to-tradable output ratio is positively related to the movement of the relative price of non-tradable goods.
10 2.7- Schooling System in Kenya
The current system of education was established in 1985. It follows the USA educational system model; the basic (primary and middle) education consists of 8 years followed by 4 years of secondary education and 4 years of university education. The average years of schooling in Kenya was 6.3 years10 in 2015, which means that on average the majority of Kenyans did not complete
their basic education. In 201011, 55% of Kenyans aged 15 and above had attended primary school
but only 22% had completed it; 22.5% had attended secondary school but 13.3% had completed it; and 6.5% had attended tertiary schooling, but with a completion rate of only 3.4%. Government spending on public education was on average 5.36% of the total GDP for the 1970–2014 period and 5.45% of the total GDP during the 2002–2014 period.
2.8- Foreign Aid
Kenya is a typical aid recipient country that receives a considerable amount of foreign aid. Aid flows to Kenya (in 2010 constant USD) have been volatile, increasing from US $242 million in 1970 (3.6% of GDP) to a peak of US $3.5 billion in 1993 (16% of GDP - see Figure 2.2). However, foreign aid dropped during the 1994–1999 period due to actions that were taken by international donors, the International Monetary Fund (IMF) and the World Bank, who forced the Kenyan government to start economic reforms in 1994. Foreign aid reached a low of $630 million (2.4% of GDP) in 1999 with some recovery thereafter in response to the devastating drought that hit Northern Kenya in 2000 where 3 million people starved. The resumption of aid coincided with the change of the Kenyan government after the 2002 elections.
10 Education Index issued by UN. 11 See Baroo and Lee (2013).
0 50 100 150 0 0.5 1 1.5 2 2.5 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 Re la tiv e P rice Ra tio Years
Figure 2.1: Non-tradable-to-tradable output ratio and the relative price of non-tradable goods (Source: Author's calculation based on WDI, 2017)
11 Figure 2.2: Kenyan aid-to-GDP ratio 1970-2014 (Source: WDI, 2017)
2 4 6 8 10 12 14 16 18 1970 1975 1980 1985 1990 1995 2000 2005 2010 Aid/GDP
From 2002 till 2014 aid-to-GDP averaged around 5% and aid to education as a total percentage of total aid disbursed did not exceed 5%, that is only 0.25% of total aid went directly to support the education system in Kenya. Figure 2.3 illustrates the distribution of aid into education among different educational levels for the 2002-2014 period. The primary education received 42%, secondary education received 12%, tertiary education received 23% of total aid to education, and the rest were used for unspecified purposes.
Primary, 42%
Secondary, 12% Tertiary, 23%
Unspecified, 23%
FIGURE 2.3: AID TO EDUCATION SHARE BY EDUCATIONAL LEVELS(AVERAGES 2002-2014. SOURCE: OECD, 2017)
12
Chapter 3
Have Sub-Saharan African Countries Caught the
“Dutch Disease” from Foreign Aid?
A Cointegrated VAR Analysis
3.1-Introduction
The literature on the effectiveness of foreign aid has widely diverse results and views (e.g. Howes (2011)). In a recent paper, Juselius et al. (2014) discuss the methodological issues behind the widely varying results in the literature. The use of a single equation to estimate the effect of foreign aid usually suffers from endogeneity bias; a valid instrumental variable for foreign aid is hard to find. In addition, the use of cross-country data analysis does not capture the dynamic effects of aid and its short-run and long-run effects on the macro economy. Durlauf (2002) argues that cross-country growth regressions are not informative for policymakers as there is no relationship between statistical significance and policy implications. Furthermore, panel data models are estimated under strict assumptions about the causal mechanisms. In contrast to these approaches, the unrestricted country-specific cointegrated vector autoregressive (CVAR) methodology allows for the identification of long-run results, without imposing strong statistical priors while controlling for country-specific characteristics and extreme events.
Juselius et al. (2014) apply their CVAR methodology to examine SSA countries, as this region is the leading case for identifying whether foreign aid is effective. They provide an overview of the long-run impact of foreign aid on real GDP, investment, private consumption, and government expenditure in 36 SSA countries. They find that in 27 SSA countries, foreign aid has a positive impact on either GDP or investment, or on both. The findings of Juselius et al. (2014) are compelling because they employ a CVAR methodology and control for a number of country-specific characteristics including extraordinary events. However, they use PPP adjusted data which precludes the change in the relative prices within the country.
13 I extend the Juselius et al. (2014) model to examine the impact of aid on GDP and its composition into tradable and non-tradable sectors. In addition to being more disaggregated, my approach allows for the identification of the “Dutch Disease”. The “Dutch Disease” occurs when a positive shock to the foreign aid reduces GDP by decreasing the relative price of tradable to non-tradable goods, and reducing the size of the tradable sector.
The term “Dutch Disease” was coined by The Economist in 1977; Corden (1984) describes the situation in the Netherlands following the discovery of natural gas in the 1960s. The appreciation of the Dutch Guilder following the discovery reduced the competitiveness of manufactured goods exporters. In the same way, foreign aid in less-developed countries may negatively impact those sectors most in need of development, such as the tradable sector (Larrain and Sachs, 1993). This chapter provides a comprehensive analysis of the long-run impact of foreign aid on GDP, tradable and non-tradable outputs, and tradable and non-tradable goods prices. Based on the reliable data for the tradable and non-tradable sectors for 22 SSA countries, during a period spanning from the mid-1960s to 2014, I contribute to the literature in three ways.
First, I develop a CVAR model designed to examine the long-run impact of aid flows on GDP and its composition for each country. I draw several conclusions motivated by the presence of two types of heterogeneity: aid heterogeneity, where each country receives a different aid-to-GDP share; and GDP’s composition heterogeneity, where GDP’s structure of tradable and non-tradable sectors differs across countries. I find that aid has a persistent negative long-run impact on GDP in eight countries, a positive impact in three countries, and no effect on the rest of the countries. Foreign aid also has diverse impacts on GDP’s structure. I find that foreign aid has a positive impact on the tradable output in four countries and on the non-tradable output in six countries. On the other hand, aid flows have persistent negative impacts on the tradable and non-tradable sectors in 16 and five countries, respectively.
Second, the analysis does not support that foreign aid to SSA countries causes the “Dutch Disease” in the long-run. I do not find that negative impacts on the tradable sector are caused by the mechanism of a decrease in the relative price of tradable to non-tradable goods12. This raises a
future research question of what transmission mechanism in the SSA countries causes the tradable sectors to shrink when aid increases.
12 No single country has experience the four phases of “Dutch Disease”: decrease in the price of tradable
goods, increase in the price of non-tradable goods, decrease in the tradable output, and decrease in total output.
14 Third, I examine whether country’s macro variables impact the amount of foreign aid a country receives. I find that GDP, tradable output, and tradable and non-tradable goods prices do affect the amount of foreign aid a country receives: aid is found to be neither weakly exogenous nor completely endogenous in 15 countries. These variables have no impact on foreign aid in six countries: aid is considered as weakly exogenous.
In my model, I take into account a number of variables and factors that reflect country-specific characteristics. Some idiosyncratic events have a significant impact on the economy. For instance, structural reforms, civil wars, conflicts, droughts, floods, and policy interventions may impact the variables of interest in ways which may change the results (Nielsen, 2008). These variables tend to bias the model parameter estimates unless adequately controlled for. In addition to permanent and transitory impulse dummies to account for extraordinary shocks, I use step dummies to account for equilibrium mean shifts in the equilibrium long-run relations.
The impact of aid on GDP and its composition is also examined by Arndt et al. (2015), Kumi et al. (2017), and Selaya and Thiele (2010). My paper differs in five key ways. First, I investigate the long-run impact of aid on the tradable and non-tradable sectors, whereas these papers analyze the impact of aid on the main sectors (agriculture, manufacturing, industry, and services sectors). Second, I use aid in levels, whereas aid as a share of GDP is used in the other papers. Unlike aid-to-GDP ratio in some countries, aid in levels is non-stationary. Third, the econometric methodology is different: I use a time series approach for each country similar to the Juselius et al. (2014) methodology. Arndt et al. (2015) use cross-section data analysis, and the other two papers use a generalized method of moments (GMM-SYS). Fourth, heterogeneity is more robustly addressed in my model, where each country is modeled with its idiosyncratic characteristics. Fifth, for policy implications, it is preferable to separately study the impact of aid on each country’s economy, since one policy that works well in one country may not necessarily work in another.
Understanding the roles of the tradable sector is important for development policy. In 2015, the United Nations adopted a set of seventeen Sustainable Development Goals (SDGs). Goal 2 is to “end hunger, achieve food security and improved nutrition, and promote sustainable agriculture” by 2030; Goal 9 is to “build resilient infrastructure, promote inclusive and sustainable industrialization, and foster innovation”.13 These objectives could be better achieved if
less-developed countries were able to increase agricultural productivity, manufacturing productivity, and the incomes of small-scale food producers. The tradable sector is composed of agricultural and
15 manufacturing sectors and is important for several reasons. First, the tradable sector in less-developed countries is labor intensive (Arellano et al., 2009; Rajan and Subramanian, 2006; 2008), and any change in its size would have a direct impact on the return to labor. Second, paying back foreign debt to the lenders involves tradable goods and is problematic when the tradable sector is shrinking. Third, exports of tradable goods are beneficial for economic growth.
The remainder of this chapter is organized as follows. Section 3.2 reviews the literature. Section 3.3 introduces the key variables and discusses the data sources and stationarity testing. The econometric approach is discussed in Section 3.4. Section 3.5 discusses the causal links between foreign aid and other variables. Section 3.6 shows the results of my analysis of the impact of foreign aid on the key variables. Robustness checking is presented in Section 3.7. Section 3.8 examines the impact of GDP and the tradable output on the flows of foreign aid. Section 3.9 summarizes and concludes.
3.2- Literature Review
The empirical literature on aid effectiveness on economic growth has not yielded clear and unambiguous results due to “heterogeneity of aid motives, the limitations of the tools of analysis, and the complex causality chain linking external aid to final outcomes” (Bourguignon and Sundberg, 2007). A number of studies find a positive effect of foreign aid on economic growth (e.g. Collier and Hoeffler, 2004; Dalgaard et al., 2004; Feeny and McGillivary, 2010; Hansen and Tarp, 2000; Karras, 2006; Kosack and Tobin, 2006; Loxley and Sackery, 2008; Juselius et al., 2014). On the other hand, there exist a number of studies that find little or no impact of aid on economic growth (e.g. Boone, 1996; Brumm, 2003; Easterly et al., 2004; Markandya et al., 2010; Ovaska, 2003; Rajan and Subramanian, 2008).
Researchers have been trying to answer if foreign aid is effective or not for recipient countries economic growth and development, but there is no definitive answer. Howes (2011) has classified the conflicting views on aid effectiveness into four categories depending on two dimensions — good or bad (outcome kind) and large or small (outcome scale) —as follows. First, good and large: Sachs (2005) argues that foreign aid could have a transformative effect on growth if applied in the right way and large volumes. Second, bad and large: Bauer and Yamey (1982) argue that aid has a negative large impact. Third, good and small: foreign aid has a minor and positive effect on development (Birdsall et al., 2005). Fourth, bad and small: aid has a little and negative effect on development (Easterly, 2006).
16 According to Feeny and McGillivary (2010), foreign aid spurred economic growth in recipient countries and it works better in some countries than in others. They find that foreign aid is found to have diminishing returns and specific types of aid are likely to be effective in poverty reduction and on economic growth. On the other hand, Rajan and Subramanian (2008) find little robust evidence of a positive or negative relationship between foreign aid and economic growth by using cross-sectional and panel data from 107 countries for a period spanning from 1960 till 2000. They also find no evidence that aid works better in countries adopting better policies, contradicting Burnside and Dollar (2000) who argue that aid is effective in a good policy environment.
3.2.1- Foreign Aid and the GDP’s Composition
Arndt et al. (2015) study the effect of foreign aid on economic transformation for 78 countries for the 1970–2007 period and show that aid reduces poverty and stimulates growth. They also find that these countries experience economic transformation as the agricultural sector as a share of GDP is shrinking, while industrial and services sectors as a share of GDP are expanding. Their results coincide with Arndt et al. (2010) in a similar study but for a shorter period (1970–2000).
In a recent study, Kumi et al. (2017) examine the impact of aid and its volatility on agricultural, manufacturing, and services sectors in 37 SSA countries for the period 1980–2014 using a GMM-SYS. They find a positive and significant effect of aid on agricultural, manufacturing, and service sectors suggesting that aid inflows spur both the tradable14 and non-tradable sectors. They find that
the volatility of aid inflows has a negative impact on the manufacturing and services sectors. On the other hand, Arellano et al. (2009) find that aid inflows, to 73 aid-receiving countries for a period spanning from 1981 to 2000, have a negative effect on the manufacturing exports. Their results suggest that an additional 1% increase in the aid-to-GDP ratio is associated with a decrease in the manufactured exports by about 0.3-0.5% of total exports.
Selaya and Thiele (2010) examine the effect of aid on GDP’s sectoral growth in 65 countries using a GMM-SYS regressions over a period spanning from 1962 to 2001. They disaggregate GDP into three parts: agriculture, industry, and services; they estimate the role of aid as a share of GDP on each sectoral growth rate. Their main findings suggest that aid has a positive effect on economic growth and on the tradable15 and non-tradable growth sectors. Their results show no evidence of
the “Dutch Disease” type effects. Demekas et al. (2002) propose that post-conflict reconstruction aid may raise the capital stock equilibrium level and does not necessarily lead to a shrinkage of the
14 They assume agriculture and manufacturing sectors as tradable sectors. 15 They assume agriculture and industry sectors as tradable sectors.
17 tradable sector. In addition, Collier and Hoeffler (2004) find that aid is effective if introduced between three to seven years after the end of the war.
3.2.2- Foreign Aid and the “Dutch Disease”
There is a small empirical literature on the effect of foreign aid on the real exchange rate16 (RER)
and on the tradable output. Some studies have found that foreign aid leads to an appreciation of the real exchange rate while others have found the opposite. Rajan and Subramanian (2006) examine the impact of aid on the recipient countries and find no long-term effects on growth even with countries that adopt good policies. They use data for 32 developing countries and report that aid inflows have adverse effects on country’s competitiveness, “as reflected in a decline in the share of labor intensive and tradable industries in the manufacturing sector”. They also show that high-labor share industries grow slower than low-high-labor share industries in the developing countries that receive a substantial amount of aid. Poor countries are labor abundant and depend on exporting labor-intensive and low-skill technologies. Thus, aid inflows would have negative effects by slowing the trade that is essential to promote growth (Michaeely, 1981, Rajan and Subramanian, 2008). Moreover, Rajan and Subramanian (2011) find that aid inflows have negative implications on the country’s competitiveness reflected by the lower growth rate of exportable industries due to real exchange rate appreciation caused by foreign aid.
Yano and Nugent (1999) observe that aid is associated with the real exchange rate appreciation in about half of 44 aid-dependent countries between 1970 and 1990, and find the reverse for the other half. Elbadawi (1999) finds that aid flows have caused substantial real exchange rate appreciation in many African and non-African countries; the real exchange rate appreciation hinders export expansion. He uses data for a panel of 62 developing countries (28 from Africa). Fielding and Gibson (2013) use data for 26 African countries and estimate a VAR of three endogenous variables (GDP, RER, and GDP deflator growth rate), in addition to aid which was considered as a weakly exogenous variable. Their main conclusion suggests a presence of a real exchange rate appreciation in most SSA countries and the size of effects differ from one country to another. In contrast, they only find one country exhibiting real exchange rate depreciation.
Using a VAR model, Addison and Baliamoune-Lutz (2013) find evidence that aid causes real exchange rate appreciation in Morocco in the long run, but has no effect on the real exchange rate
16 The real exchange rate is defined by Mendoza and Uribe (2000) as the relative price of non-tradable goods
in terms of tradable goods (i.e. 𝑃𝑃𝑁𝑁
𝑃𝑃𝑇𝑇). Other studies use a proxy for the real exchange rate such as the relative
18 in Tunisia. Lartey (2007) observes that an increase in official aid causes a real appreciation of the exchange rate and its magnitude is greater than that associated with foreign direct investments (FDI) in a sample of SSA countries from a period spanning from 1980 till 2000. In a recent study, Juselius et al. (2017) find that a shock to foreign aid causes the real exchange rate to appreciate in the long-run in Ghana, but not in Tanzania. Kang et al. (2013) study the effect of aid shocks on GDP per capita growth, exports, and imports in 30 aid-receiving countries, using a heterogeneous panel VAR model. They find that in 16 countries, global aid has a negative effect on GDP per capita growth, consistent with the “Dutch Disease” hypothesis.
On the other hand, Li and Rowe (2007) find, by using a vector error correction model (VECM), that increases in aid inflows are associated with a depreciation in the real effective exchange rate in Tanzania from 1970 till 2005. Issa and Ouattara (2008) find, by using autoregressive distributed lag (ARDL) model, no evidence in the short-run or in the long-run of the “Dutch Disease” in Syria during the 1965–1997 period. Mongardini and Rayner (2009) use a pooled mean group estimator for 24 SSA countries from 1980 to 2006. They find that grants and remittances are not associated with an appreciation of the real exchange rate in SSA countries in the long run, therefore no “Dutch Disease” effects.
The existing empirical papers estimate the real exchange rate indirectly; while the present chapter directly constructs the real exchange rate from tradable and non-tradable goods prices. In contrast to most of the existing literature which uses cross-country or panel regressions, my approach is based on a single country time-series analysis where 22 CVAR models are estimated. This chapter extends the above empirical literature by employing a multivariate time-series approach to investigate the long-run impact of foreign aid on GDP, tradable and non-tradable outputs, and tradable and non-tradable goods prices.
19 3.3- Data Sources and Stationarity Tests
3.3.1- Data Sources
Data for GDP, tradable and non-tradable outputs in current and constant value-added prices are taken from the World Development Indicators (WDI, 2017)17. Official Development Aid (ODA)
data18 are sourced from the Organisation for Economic Co-operation and Development (OECD).
Most African countries have big informal and subsistence sectors and even in the formal sectors not all the transactions are formally recorded (Jerven, 2010). The international agencies (The World Bank, IMF, OECD, etc…) supply aggregate national data and apply different statistical methods to gather them into continuous series over different base years. The main sources of publicly available data are the WDI, IMF, and Penn Tables (PWT); using different data sources and data transformation may lead to different results (Juselius et al., 2014). The WDI database are better suited for this study as the WDI is a reliable source at disaggregated level data in constant and current prices of agricultural, manufacturing, industrial, and services value-added.
Table 3.1A, presented at the end of the chapter, shows the SSA countries that are included in this study. Out of the 48 SSA countries only 22 countries are included. The rest are excluded (See Table A1, Appendix A.1) due to many reasons such as missing data on the tradable and non-tradable outputs, negative aid (Gabon and Mauritius in 2003), stationarity of aid time-series data (Comoros, Congo Republic, Namibia, and Swaziland), and prices (prices are found to be I(2) in Zambia). To mitigate the problem of degrees of freedom erosion in the CVAR model, I make the sample size as big as possible. The availability of data constraints the sample size to a minimum of 33 annual observations for Madagascar to a maximum of 51 for Sierra Leone.
The heterogeneity of aid-to-GDP and the heterogeneity of tradable output-to-GDP ratios are addressed in this empirical study. The aid-to-GDP ratio ranges from 4% in Cameroon to 25.3% in Cape Verde with an average of 12% for the overall sample. When agricultural and manufacturing sectors constitute the tradable sector, tradable output-to-GDP ratio ranges between 18.3% in Botswana to 53.65% in Central African Republic. When the tradable output is composed of agricultural and industrial sectors, tradable output-to-GDP ranges from 32.8% in Cape Verde to 61% in Sierra Leone. Kenya’s aid-to-GDP ratio equals 5.75% (half of the sample), while Kenya’s tradable output constitute 38% of GDP.
17 Downloaded on 15/09/2017.
18 Foreign aid data disaggregated at sectorial level starts from 2002. This short period considers inapplicable
20 The data on exports and imports as a share of GDP are presented in Table 3.1B. This table gives a perspective on how dependent each economy is on exporting and importing goods and services. Exports-to-GDP ranges from 2.65% in Ethiopia to a maximum of 49% in Botswana. Whereas, imports-to-GDP ranges between 9.4% in Ethiopia to a maximum of 74% in Seychelles. All SSA countries exhibit a trade deficit that ranges from 3.2% in Botswana to 47% in Lesotho. Furthermore, aid-to-GDP exceeds exports-to-GDP in five SSA countries19; whereas, in 12 SSA countries20
aid-to-GDP ratio exceeds the trade deficit. These numbers emphasize the importance of aid to SSA countries as a source of income.
3.3.1.1 Human Capital Data
For this empirical analysis, I have not included a relevant variable to measure human capital such as educational attainment or enrollment for the following reasons:
a. The data on the average years of schooling are available only every 5 years (Barro and Lee, 2013). These data are not continuous.
b. The data on the education index which is one of three components of the UN Human Development Index starts from 1990.
c. The number of pupils attending primary, secondary, and tertiary education has a lot of missing data.
d. The human capital index which is introduced by PWT version 8 and is “based on the average years of schooling from Barro and Lee (2013) and an assumed rate of return to education, based on Mincer equation estimates around the world (Psacharopoulos, 1994)”. These data seem to be I(2) and cannot be used in the present model.
3.3.2- Data Transformation
The data are annual observations and comprise of the following variables: net Official Development Assistance (𝑎𝑎𝑎𝑎𝑎𝑎𝑡𝑡), Gross Domestic Product (𝑔𝑔𝑎𝑎𝑔𝑔𝑡𝑡), tradable output (𝑡𝑡𝑡𝑡𝑎𝑎𝑎𝑎𝑡𝑡), non-tradable output (𝑛𝑛𝑡𝑡𝑡𝑡𝑎𝑎𝑎𝑎𝑡𝑡), price of the tradable goods ( 𝑔𝑔𝑡𝑡𝑇𝑇), and price of the non-tradable goods ( 𝑔𝑔𝑡𝑡𝑁𝑁𝑇𝑇). Small letters denote logarithmic values. Macroeconomic variables are usually trending over time, suggesting a multiplicative rather than an additive specification; by taking logs an additive model form can be
19 Burkina Faso, Burundi, Ethiopia, Lesotho, and Rwanda.
20 Botswana, Burundi, Cameroon, Central Africa, Ethiopia, Gambia, Malawi, Mali, Mauritania, Rwanda,
21 achieved. I follow Juselius et al. (2014) logarithmic transformation of the variables at levels instead of using ratios such as aid-to-GDP and tradable output-to-GDP as these ratios are usually bounded between 0% and 100%. In addition, I do not use GDP per capita as the logarithm of population is usually found close to I(2), in contrast to logarithm of real GDP which is found to be I(1) (Juselius et al., 2014). Dividing GDP (I(1) variable) with population (I(2) variable) is likely to weaken any statistical inferences as Kongsted (2005) suggests. Data definitions and measurements are presented in Appendix A.4.
3.3.3-Stationarity Tests
Time-series data are time dependent and it is crucial that the data are investigated and tested if they are stationary21, non-stationary22, and whether the series are subject to structural breaks. It is
preferable to start with graphical representations of the level and first difference of the series to reveal data features23. The variables’ series (𝑎𝑎𝑎𝑎𝑎𝑎
𝑡𝑡, 𝑔𝑔𝑎𝑎𝑔𝑔𝑡𝑡, 𝑛𝑛𝑡𝑡𝑡𝑡𝑎𝑎𝑎𝑎𝑡𝑡, 𝑡𝑡𝑡𝑡𝑎𝑎𝑎𝑎𝑡𝑡, 𝑔𝑔𝑡𝑡𝑇𝑇 and 𝑔𝑔𝑡𝑡𝑁𝑁𝑇𝑇)24 in levels
exhibit trend like behavior over the sample period (i.e. trending) and the first difference appears stationary around the trend (i.e. trend-stationary).
Time series data are in general non-stationary data and cross estimations can result in spurious results that lead to meaningless statistical results. To get credible results, time series data should be differenced d times to become stationary; that is, the series is integrated of order d and denoted by I(d).
3.3.3.1- Unit Root Tests25
I begin with a preliminary assessment of the presence of unit roots in the data by conducting several unit-root tests. In this study, the time-series property of each variable is investigated through the Augmented Dickey-Fuller (ADF), the Philipps-Perron (PP), and the Kwiatkowski-Phillips-Schmidt-Shin (KPSS) tests for the unit root26.
The ADF (1979) tests the null hypothesis that the series has a unit root versus the alternative hypothesis that the series is stationary or trend stationary. If the test t-statistic is less than the critical value (negative), then the null hypothesis can be rejected and no unit root is present in the series.
21 Mean is constant; variance is finite and does not depend on time. 22 Mean and variance vary over time.
23 The graphs are not presented in this paper and can be presented upon request.
24 The relative price of non-tradable good is stationary in most countries; as it is a ratio of 2 ratios: 𝑔𝑔𝑁𝑁= 𝑝𝑝𝑁𝑁𝑇𝑇
𝑝𝑝𝑇𝑇 25 I have relied on testing individual series and not rely only on rank testing alone as some variables are I(2)
such as the prices in Zambia (excluded from the study).
